Inspiration
“Wooden Letters to Home” draws inspiration from one of the earliest known personal letters in Chinese history — the Heifu Wooden Slips, discovered in Hubei Province. These wooden slips recorded the thoughts of an ordinary Qin soldier writing home during the late Warring States period. Amid the vastness of imperial conquest and the brutality of the Qin–Chu war, we chose to magnify the life of one insignificant man — a soldier who writes, hopes, fears, and loves. Rather than glorifying warfare, the film attempts to restore humanity within history: the trembling handwriting, the longing for spring, the fragile warmth between brothers, and the unbearable silence between truth and survival.
What it does
The film follows Heifu, a Qin soldier marching into the Qin–Chu war, who writes letters home while battling fear, cold, and the brutality of the battlefield. Through intimate fragments — campfire conversations, military orders at dawn, old soldiers’ memories, and the final tragedy — the short film reveals how war silently consumes youth, family ties, and promises. The story blurs truth and lies: after his younger brother Jing dies in battle, Heifu continues writing “they are both alive.” What he sends home is not the truth, but the last warmth he can keep alive.
How we built it
The film integrates AIGC image generation with traditional cinematic language. Ideation & Scriptwriting: Story structure and emotion arcs were developed using conventional film-writing methods. Image Generation: YiDream AI was used to create initial stills, with prompt engineering ensuring historical accuracy in costumes, props, and environments. Character Consistency: ComfyUI face-transfer workflows were employed to maintain the same characters across scenes. Important props — like the talisman — were embedded through style-transfer in Kontext & ComfyUI. Visual Refinement: Magnific enhanced key frames to reach cinematic clarity. Dynamic Shots: Vidu, Keling, and YiDream video tools produced movement, lip-syncs, and atmospheric sequences. Editing & Sound: Video was cut on Jianying; music was generated via Suno AI; sound effects combined AI tools with manual adjustments. The entire workflow forms a hybrid pipeline of “AI generation → human calibration → cinematic post-production.”
Challenges we ran into
Maintaining historical accuracy while using generative models trained on mixed datasets. Ensuring character consistency across more than 70 shots, requiring repeated re-generation and manual correction. Balancing AI randomness with strict cinematic storytelling — making sure the film feels coherent, not algorithmic. Rendering battle scenes that achieve emotional weight without becoming visually chaotic. Reconciling AI’s aesthetics with the film’s intended restraint and solemn tone.
Accomplishments that we're proud of
We recreated a lost fragment of ancient history through AIGC, making invisible emotions visible. Achieved a stable character continuity across complex, multi-tool workflows. Built a visual world inspired by archaeological relics rather than fantasy tropes. Formed a collaborative pipeline where AI amplifies — rather than replaces — human creativity. Told a war story without glorification: gentle, restrained, human.
What we learned
We learned that AI can assist historical filmmaking, but it cannot replace the emotional precision that comes from human judgment. The film demonstrated that with the right constraints and direction, AIGC can support narrative cinema — not as a shortcut, but as a new form of craftsmanship. We also learned that the smallest stories in history often have the largest emotional resonance.
What's next for Wooden Letters to Home
Expand the story into a longer short film exploring the Qin–Chu war from multiple civilian perspectives. Create an interactive digital exhibit combining the original Heifu Wooden Slips with AI-generated imagery. Develop a series on “lost voices in ancient history,” using AIGC to revive unheard personal stories. Continue refining our AI–human hybrid pipeline to push cinematic quality further.
Built With
- comfyui
- kling
- seedream
- vidu
Log in or sign up for Devpost to join the conversation.